|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace App\Model; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* A Query produces the parameters needed to be passed to the CloudElastic service. |
|
9
|
|
|
*/ |
|
10
|
|
|
class Query |
|
11
|
|
|
{ |
|
12
|
|
|
public const PRE_TAG = '%**%'; |
|
13
|
|
|
public const POST_TAG = '*%%*'; |
|
14
|
|
|
public const MAX_RESULTS = 5000; |
|
15
|
|
|
|
|
16
|
|
|
/** @var string The query string. */ |
|
17
|
|
|
protected $query; |
|
18
|
|
|
|
|
19
|
|
|
/** @var int[] Array of namespace IDs. */ |
|
20
|
|
|
protected $namespaces; |
|
21
|
|
|
|
|
22
|
|
|
/** @var bool Whether to get params for a regular expression search. */ |
|
23
|
|
|
protected $regex; |
|
24
|
|
|
|
|
25
|
|
|
/** @var bool Whether the params should be for a case-insensitive search. */ |
|
26
|
|
|
protected $ignoreCase; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* Query constructor. |
|
30
|
|
|
* @param string $query |
|
31
|
|
|
* @param int[] $namespaces |
|
32
|
|
|
* @param bool $regex |
|
33
|
|
|
* @param bool $ignoreCase |
|
34
|
|
|
*/ |
|
35
|
|
|
public function __construct(string $query, array $namespaces, bool $regex = false, bool $ignoreCase = false) |
|
36
|
|
|
{ |
|
37
|
|
|
// Silently use regex to do exact match if query is wrapped in double-quotes. |
|
38
|
|
|
if ('"' === substr($query, 0, 1) && '"' === substr($query, -1, 1)) { |
|
39
|
|
|
$regex = true; |
|
40
|
|
|
$query = preg_quote(substr($query, 1, -1)); |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
$this->query = $query; |
|
44
|
|
|
$this->namespaces = $namespaces; |
|
45
|
|
|
$this->regex = $regex; |
|
46
|
|
|
$this->ignoreCase = $ignoreCase; |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
/** |
|
50
|
|
|
* Get parameters needed to make the CloudElastic query. |
|
51
|
|
|
* @return array|mixed[] |
|
52
|
|
|
*/ |
|
53
|
|
|
public function getParams(): array |
|
54
|
|
|
{ |
|
55
|
|
|
$params = $this->regex ? $this->getRegexQuery() : $this->getPlainQuery(); |
|
56
|
|
|
|
|
57
|
|
|
if (!empty($this->namespaces)) { |
|
58
|
|
|
$params['query']['bool']['filter'][] = [ 'terms' => [ |
|
59
|
|
|
'namespace' => $this->namespaces, |
|
60
|
|
|
] ]; |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
return $params; |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
/** |
|
67
|
|
|
* Params to be passed to CloudElastic for a plain (normal) query. |
|
68
|
|
|
* @return mixed[] |
|
69
|
|
|
*/ |
|
70
|
|
|
private function getPlainQuery(): array |
|
71
|
|
|
{ |
|
72
|
|
|
return [ |
|
73
|
|
|
'timeout' => '150s', |
|
74
|
|
|
'size' => self::MAX_RESULTS, |
|
75
|
|
|
'_source' => ['wiki', 'namespace_text', 'title'], |
|
76
|
|
|
'query' => [ |
|
77
|
|
|
'bool' => [ |
|
78
|
|
|
'filter' => [ |
|
79
|
|
|
[ 'match' => [ |
|
80
|
|
|
'source_text.plain' => $this->query, |
|
81
|
|
|
] ], |
|
82
|
|
|
], |
|
83
|
|
|
], |
|
84
|
|
|
], |
|
85
|
|
|
'stats' => ['global-search'], |
|
86
|
|
|
'highlight' => [ |
|
87
|
|
|
'pre_tags' => [self::PRE_TAG], |
|
88
|
|
|
'post_tags' => [self::POST_TAG], |
|
89
|
|
|
'fields' => [ |
|
90
|
|
|
'source_text.plain' => [ |
|
91
|
|
|
'type' => 'experimental', |
|
92
|
|
|
], |
|
93
|
|
|
], |
|
94
|
|
|
'highlight_query' => [ |
|
95
|
|
|
'match' => [ |
|
96
|
|
|
'source_text.plain' => $this->query, |
|
97
|
|
|
], |
|
98
|
|
|
], |
|
99
|
|
|
], |
|
100
|
|
|
]; |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
/** |
|
104
|
|
|
* Params to be passed to CloudElastic for a regular expression query. |
|
105
|
|
|
* @return mixed[] |
|
106
|
|
|
*/ |
|
107
|
|
|
private function getRegexQuery(): array |
|
108
|
|
|
{ |
|
109
|
|
|
return [ |
|
110
|
|
|
'timeout' => '150s', |
|
111
|
|
|
'size' => 5000, |
|
112
|
|
|
'_source' => ['wiki', 'namespace_text', 'title'], |
|
113
|
|
|
'query' => [ |
|
114
|
|
|
'bool' => [ |
|
115
|
|
|
'filter' => [ |
|
116
|
|
|
[ 'source_regex' => [ |
|
117
|
|
|
'regex' => $this->query, |
|
118
|
|
|
'field' => 'source_text', |
|
119
|
|
|
'ngram_field' => 'source_text.trigram', |
|
120
|
|
|
'max_determinized_states' => 20000, |
|
121
|
|
|
'max_expand' => 10, |
|
122
|
|
|
'case_sensitive' => !$this->ignoreCase, |
|
123
|
|
|
'locale' => 'en', |
|
124
|
|
|
] ], |
|
125
|
|
|
], |
|
126
|
|
|
], |
|
127
|
|
|
], |
|
128
|
|
|
'stats' => ['global-search'], |
|
129
|
|
|
'highlight' => [ |
|
130
|
|
|
'pre_tags' => [self::PRE_TAG], |
|
131
|
|
|
'post_tags' => [self::POST_TAG], |
|
132
|
|
|
'fields' => [ |
|
133
|
|
|
'source_text.plain' => [ |
|
134
|
|
|
'type' => 'experimental', |
|
135
|
|
|
'number_of_fragments' => 1, |
|
136
|
|
|
'fragmenter' => 'scan', |
|
137
|
|
|
'fragment_size' => 150, |
|
138
|
|
|
'options' => [ |
|
139
|
|
|
'regex' => [$this->query], |
|
140
|
|
|
'locale' => 'en', |
|
141
|
|
|
'regex_flavor' => 'lucene', |
|
142
|
|
|
'skip_query' => true, |
|
143
|
|
|
'regex_case_insensitive' => $this->ignoreCase, |
|
144
|
|
|
'max_determinized_states' => 20000, |
|
145
|
|
|
], |
|
146
|
|
|
], |
|
147
|
|
|
], |
|
148
|
|
|
], |
|
149
|
|
|
]; |
|
150
|
|
|
} |
|
151
|
|
|
} |
|
152
|
|
|
|
MusikAnimal /
global-search
